Return on investment is the simplest measure there is, and it is silent about time. Two investments with the same ROI can be wildly different propositions.
9 min read · Level 4
ROI is usually dismissed as the crude measure that serious people graduate from, and that framing costs more than it saves. Its crudeness is also its integrity: it contains no assumption about reinvestment, no discount rate you chose, no smoothing, and no judgement about what a future cash flow is worth. Every more sophisticated measure in this subject buys its extra information by importing an assumption, and each of those assumptions is a place where a number can be made to say what its author wanted. ROI cannot be manipulated except by leaving cash movements out - which is a lie about the facts rather than a defensible modelling choice, and is therefore far easier to catch. The reasonable conclusion is not that ROI is the best measure; it is that the sophistication of a return figure and its reliability move in opposite directions, and a reader who is shown a 34% IRR should want to see the cash flows more, not less, than one who is shown a 38% ROI.
Return on investment is what you got back, less what you put in, divided by what you put in. Nothing else.
ROI = (total proceeds - total cost) / total cost
Two details decide whether the answer means anything. Total proceeds is every pound or dollar the investment ever returned - the sale price plus any income received along the way, not just the sale. And total cost is the cost basis: the purchase price plus everything that had to be paid to acquire it. Costs on the way out come off the proceeds.
What the formula contains is complete. What it omits is time, and time is not a detail that can be added afterwards - it is the difference between an ordinary result and an excellent one.
Because ROI is the number everybody quotes, in conversation, in marketing material and in their own memory of how an investment went. 'I made 38% on that' is a true sentence that describes two completely different events, and the version people remember is always the more flattering one.
And because it is the right tool in one specific situation, which is comparing two things that happened over the same period. Within a single year, across a single project, between two positions held for the same span, ROI is exact and requires no assumptions at all. Its weakness only appears the moment the holding periods differ, which is most of the time and is precisely when it is reached for.
Educational, not personalised financial advice. Nothing in the Investor Academy is a recommendation to buy, sell or hold anything — what is right for you depends on your own circumstances and the rules where you live.
Buy an asset for 50,000. While holding it, receive 4,000 of income. Sell it for 65,000.
Total proceeds: 65,000 + 4,000 = 69,000 Gain: 69,000 - 50,000 = 19,000 ROI: 19,000 / 50,000 = 38.0%
That number is correct and, on its own, close to useless. Two versions:
If the holding period was seven years, the annualised return is 1.38 to the power of one-seventh, minus one, which is 4.71% a year.
If the holding period was eighteen months, the annualised return is 1.38 to the power of one divided by 1.5, minus one, which is 23.95% a year.
Same 38%. Same 19,000 of money. One is a modest result and the other is an exceptional one, and the ROI figure cannot tell them apart because time never entered the calculation.
Now the second omission, which is smaller but is where the number gets quietly overstated. Suppose acquiring the asset cost 1,500 in fees and taxes, and selling it cost 1,300.
Cost basis: 50,000 + 1,500 = 51,500 Net proceeds: 69,000 - 1,300 = 67,700 Net gain: 67,700 - 51,500 = 16,200 ROI after costs: 16,200 / 51,500 = 31.46%
The headline 38.0% became 31.46%, a difference of 6.54 percentage points, and 2,800 of cash. Over the seven-year version, the annualised figure falls from 4.71% to 3.98%.
Neither adjustment required a different method. Both required putting every real cash movement into the formula, which is the entire discipline of this subject and is skipped more often than any other step in it.
The first mistake is comparing ROIs across different holding periods. A 38% return over eighteen months and a 38% return over seven years are not comparable quantities, and putting them side by side in a list is how a slow result gets mistaken for a good one.
The second is using the purchase price rather than the cost basis. Acquisition fees, transaction taxes, legal costs and the price of getting the asset into a usable state are all part of what the investment cost, and leaving them out flatters every figure by the same silent mechanism.
The third is forgetting income. An asset that paid nothing and one that paid 4,000 along the way are different investments, and a ROI computed from the sale price alone treats them identically.
The fourth is computing ROI on the peak valuation rather than on a sale. An unrealised gain is a mark, and a mark put through the ROI formula produces a number that reads exactly like a result and is not one.
Money out to acquire, money out for costs, money in from income, money in on sale. Dates matter even if this particular formula ignores them, because the next three lessons do not, and a cash-flow list with dates is the input for all of them.
Add every cost of acquiring and preparing the asset to the denominator, and take exit costs off the numerator. This single correction moved the example from 38.0% to 31.46%.
Never write an ROI without the period beside it. '38% over seven years' is a complete statement; '38%' is an invitation to be misunderstood, including by yourself in two years' time.
Total proceeds divided by total cost, raised to the power of one over the number of years, minus one. Do it for both investments and compare those. Without this step a comparison between different holding periods is not a comparison at all.
For a single completed investment, for comparing two things held over the same period, and for any quick check where the period is identical and known.
Avoid it as the comparison tool across different holding periods, and avoid it entirely where money went in or out at several different times - a ROI on a position that was added to twice and partly sold once is arithmetic performed on a quantity that does not exist.
A consistent finding when investors are asked to reconstruct a past result from records rather than memory is that the recalled figure is both higher and shorter than the real one. The mechanism is ordinary rather than dishonest: the purchase price is remembered because it was a single decision, while acquisition costs, the year the position went nowhere, and the fees on exit were each small at the time and none of them attached themselves to the story. The remembered version is the headline and the real version is the headline minus everything that was unmemorable.
Rebuild one past return from actual records before quoting it, including to yourself. The number you remember was assembled by the part of you that was pleased at the time.
Create a free account to read the rest and apply it to your own investing.